2026
The Automation Anxiety Report
How AI Is Reshaping What US Workers Tell Employers, Coworkers, and Themselves
Download the Report69%
of US workers expect parts of their job to be automated by AI within 24 months. The figure rises to 79% for Gen Z.
63%
have lied or exaggerated their AI skills to appear more knowledgeable than they are. The figure reaches 80% for Gen Z and 70% for men.
81%
of workers report engaging in at least one behavior to discourage or limit AI use at work, while simultaneously overstating their own AI skills.
64%
of workers have not had their AI skills verified by an employer. 76% say AI misrepresentation creates business risk.
29%
of workers would present themselves more honestly if employers clearly disclosed what would be independently verified.
Executive Summary
AI is creating the next wave of workforce disruption. The Automation Anxiety Report™ 2026, based on a national survey of 1,500 US full-time employed adults, finds that the disruption is not only operational. It is a credibility crisis. Workers expect AI to reshape their jobs but do not feel fully prepared for it. In response, they are overstating their capabilities, while simultaneously resisting the AI adoption their employers are paying for. Employers, in turn, lack consistent ways to verify the new category of skill claim now appearing on resumes and in workplaces. The result is a growing gap between perceived and actual capability, with measurable risk for both employees and organizations.
“Automation anxiety is not just about job loss. It is reshaping behavior, distorting skill signals, and challenging the credibility of the modern workforce.”
-
AI anxiety is widespread and personal. 69% of workers believe AI will automate parts of their role within 24 months. The concern is grounded in direct experience: 40% of workers who expect AI impact have personally observed AI tools doing parts of their work.
-
AI skills misinformation is becoming the new normal. 63% of workers have lied or exaggerated their AI skills, with 80% of Gen Z and 70% of men reporting the behavior. Only 34% of workers who list AI skills can confidently perform all of them at a professional level.
-
The exaggeration shows up across the workplace. 40% speak confidently about AI in meetings to avoid appearing behind. 25% take credit for AI-assisted work as their own. 16% explicitly lie about having AI skills.
-
A second contradiction runs in the opposite direction. 81% of workers admit to discouraging or limiting AI use at work, including 53% who prefer manual approaches and 45% who raise risk concerns more strongly than they personally believe.
-
Verification infrastructure has not kept up. 64% of workers have not had their AI skills tested by their employer. 39% believe their employer can effectively verify AI skills. 43% believe most coworkers are exaggerating.
-
The mandate for reform is coming from the workforce itself. 76% of workers say AI misrepresentation creates business risk. 61% want human review of AI-related hiring decisions. 29% would be more honest if verification scope were clearly disclosed.
Methodology
Online quantitative survey (Pollfish instrument ID 395577040) of n=1,500 US adults employed full-time. Fielded April 21–22, 2026, over a continuous 16-hour field window. Median completion time: 4 minutes 44 seconds. Margin of error: ±2.5 percentage points at 95% confidence at full sample.
Sample composition: 56% male (n=837), 44% female (n=663). Generational breakdown: Gen Z 18–29 (n=132), Millennials 30–45 (n=651), Gen X 46–61 (n=563), Baby Boomers 62–80 (n=154). Race and ethnicity: White (n=1,033), Black or African American (n=187), Hispanic (n=133), Asian (n=90). 63% hold a Bachelor’s degree or higher. All respondents passed an AI familiarity screener; 98% reported substantive familiarity with AI tools.
Subsample sizes and margins of error. Subsamples below n=100 are reported as directional only. The Asian cohort (n=90) sits just below this threshold and is flagged accordingly. The Baby Boomer cohort (n=154) and the Gen Z cohort (n=132) carry small-sample caveats per GCheck protocol.
Subsample sizes and margins of error
| Cohort | n | MoE (95% CI) | Reporting posture |
|---|---|---|---|
| Full sample | 1,500 | ±2.5 pts | Primary |
| Millennials (30-45) | 651 | ±3.8 pts | Primary |
| Gen X (46-61) | 563 | ±4.1 pts | Primary |
| Gen Z (18-29) | 132 | ±8.5 pts | Small-sample caveat |
| Baby Boomers (62-80) | 154 | ±7.9 pts | Small-sample caveat |
| White | 1,033 | ±3.0 pts | Primary |
| Black or African American | 187 | ±7.2 pts | Primary |
| Hispanic | 133 | ±8.5 pts | Primary |
| Asian | 90 | ±10.3 pts | Directional only |
Conditional questions are reported against their conditional base rather than the full sample. Q6 (reasons for AI fluency theater) is reported against n=952 respondents who selected at least one Q5 behavior. Q8 (reasons for AI resistance) is reported against n=1,212 respondents who selected at least one Q7 behavior. Q11 (self-rated ability) is reported against n=1,062 respondents who listed at least one AI skill in Q10. Q13 (primary reason for shadow search) is reported against n=813 respondents who answered Yes to Q12. Multi-select questions allow totals exceeding 100%.
The cross-study bridge in the closing chapter references Trust in Hiring 2026. The two samples are designed as complementary populations, not matched samples. Convergence and divergence readings are valid at the population level, not the individual level. All percentages are unweighted; results reflect sentiment within the survey window and may shift with subsequent AI market developments.
AI Anxiety Is Widespread and Personal
40%
have personally observed AI tools that can already perform parts of their work
38%
cite predictions in their field about AI replacing jobs
35%
have seen company announcements about implementing AI tools
29%
say AI has already taken over some of their job responsibilities
28%
have seen tools improve to the point of matching or exceeding their own output quality
26%
have seen colleagues or peers in their industry lose roles to AI or automation
“Anxiety about AI is no longer driven by what workers read in the news. It is driven by what they see at their desks.”
The AI Skills Bubble
Self-declared inflation, by the numbers
The inflation does not require external auditing. It is self-reported in the survey. Among the 1,062 workers who list at least one AI skill on their resume or LinkedIn, only 34% say they could confidently perform all of those skills at a professional level if asked to demonstrate them today. Half say they could perform most but some are still developing or overstated. Fifteen percent admit several of their listed skills exceed their actual ability. One percent cannot confidently perform any of the skills they listed.
Figure 1. 63% of US Workers Have Lied or Exaggerated AI Skills
(%) reporting AI fluency theater
Percentage (%)
AI skills appear widely on resumes
AI-related skills now appear on the majority of US worker resumes and LinkedIn profiles. 71% of US employed workers list at least one AI skill publicly. Only 29% list none at all. The most commonly claimed skills span the full gradient from broad familiarity to specialist expertise: AI tool proficiency at 40%, AI content creation at 39%, AI-assisted data analysis at 30%, AI for productivity at 23%, machine learning at 13%, and AI product development at 15%. Specialist credentials appear at base rates that far exceed actual labor-market supply for those skills.
The AI Skills Bubble
A 2026 workforce phenomenon in which AI skill claims have outpaced workers’ actual ability to perform them, documented at industrial scale through self-reported inflation rather than inferred from external assessment.
47% of US employed workers list AI skills publicly and privately admit some or all exceed their actual ability.
Self-rated ability tells a different story
Workers themselves describe a gap between what they list and what they can do. Among the 71% who list AI skills publicly, only 34% say they could confidently perform all of those skills at a professional level if asked to demonstrate them today. Half can perform most but say some are still developing or overstated. 15% admit they have only basic familiarity, with several listed skills exceeding their actual ability. 1% cannot confidently perform any of the skills they listed.
Expressed as a share of the full workforce, 47% of US employed workers list AI skills publicly and privately acknowledge that some or all of those skills exceed their actual ability. This is the AI Skills Bubble in its most mechanically defensible form: documented in workers' own self-assessment, not inferred from external evaluation.
From Subtle Inflation to Direct Misrepresentation
The exaggeration takes multiple forms, many of which extend beyond resumes into everyday workplace behavior. The behaviors span a spectrum from soft impression management to explicit fabrication. Even at the conservative end, one in six US workers has explicitly lied to colleagues or managers about having AI skills they do not have.
40%
have spoken confidently about AI in meetings to avoid appearing behind
33%
have let others assume they have stronger AI skills than they do
31%
have exaggerated their experience with AI tools, rising to 50% for Gen Z, 38% for Millennials, 21% for Gen X, and 19% for Boomers
25%
have taken credit for AI-assisted work as entirely their own
19%
have acted as if using AI when they were not
18%
have volunteered for AI tasks they were not qualified for, rising to 30% for Gen Z and 24% for Millennials
16%
have explicitly lied about having AI skills they did not possess, rising to 29% for Gen Z and 24% for Millennials
“Even at the conservative end of the distribution, one in six US workers has explicitly lied about having AI skills they do not have.”
Why Workers Feel Compelled to Exaggerate
Among the 952 workers who report at least one AI fluency theater behavior, the motivations form a composite of skill-building intent, competitive pressure, structural training gaps, and fear. The single most common reason is anticipatory: workers who plan to learn AI eventually but have decided to claim the credential first. The next most common is normative: the belief that everyone else is doing the same thing.
76%
plan to build the skills over time, even if they are overstating now
70%
believe most people in their industry are also exaggerating their AI expertise
57%
felt pressure to appear AI-capable even if their skills are not yet there
53%
say the job market is so difficult they feared losing their job without AI skills
52%
feel they have no choice or fear being seen as replaceable if they do not adapt
52%
have not received proper training or support to actually learn AI tools
48%
do not believe their employer has a reliable way to verify AI-related skills
46%
feared they would be fired if they did not appear to have AI skills
“ Pressure to compete drives exaggeration. Weak verification allows it to continue. The perception that everyone is doing it normalizes the behavior across industries.”
The Double Distortion
53%
have preferred manual approaches specifically to avoid increasing reliance on AI
45%
have raised risk concerns about AI more strongly than they personally believe those risks
28%
have said using AI takes more time or effort than it actually does
24%
have downplayed or dismissed AI tools as "not useful" when they had in fact helped
23%
have acted as if AI outputs were inaccurate when they were acceptable
72%
are concerned that widespread AI adoption will reduce job opportunities broadly
71%
do not fully trust the quality or accuracy of AI outputs
63%
worry that reliance on AI will erode their own skills over time
57%
have not received enough training or support to use AI effectively
55%
are concerned that relying on AI could make their role easier to replace
51%
are concerned that increased AI use could reduce the need for their role
The implication for AI return on investment
The practical consequence for organizations is consequential. The Double Distortion is not a fringe behavior at this scale; it is a workforce-wide pattern. Organizations measuring AI return through surface-level adoption metrics are measuring the visible signal of AI use, not the underlying productivity gain. The suppression of that signal, conducted quietly by four in five workers, will continue to understate AI's ceiling for as long as the behavior goes unnamed and unaddressed. The workforce is not refusing AI loudly. It is slowing AI quietly, and the slow rolls into the productivity numbers.
The Double Distortion
The 2026 workforce is simultaneously overstating its AI capabilities and understating AI’s actual usefulness. Two contradictory behaviors, running in opposite directions, in the same workforce, both at majority scale.
63% have inflated their AI skills to appear more knowledgeable. 81% have engaged in behaviors to limit or discourage AI use at work.
“Workers are simultaneously overstating their ability to use AI and understating AI's actual value. The Double Distortion is the story.”
Career Movement and Workforce Instability
54%
of US employed workers are currently engaged with the job market, either actively applying or casually browsing. The number itself is a meaningful labor-market signal. The composition of the underlying motivations, in which AI-related reasons displace traditional career drivers, is more meaningful still.
26%
career advancement or higher compensation
25%
general job security anxiety, even though nothing specific has happened at their company
23%
want to move into a field they believe is more resistant to AI automation
13%
believe their current role may be automated or eliminated
9%
dissatisfaction with role, manager, or company
4%
want a different job or career path
Workers are actively trying to AI-proof themselves
Beyond shadow searching, workers are taking concrete steps to reposition themselves for an AI-shaped market. In the past six months, the share of US employed workers who have made specific resume or skill changes specifically because of AI concerns is substantial. 78% have taken at least one preemptive action.
updated their resume or LinkedIn to emphasize AI-resistant skills
added AI-related skills, including some they cannot confidently perform
started or enrolled in an AI certification program
applied to roles outside their current qualification range that seem AI-proof
inflated their technical proficiency on resume or LinkedIn
started or enrolled in a reskilling program for an AI-resistant industry
removed or downplayed parts of their experience that overlap with what AI can do
Automation anxiety is not just influencing perception. It is actively reshaping career decisions and labor-market movement. The workers who are leaving are leaving because they believe their current trajectory runs into AI displacement. An effective retention response requires more than compensation adjustment. It requires honest communication about which roles are genuinely at risk, real investment in AI skill development for workers who want to stay, and verification infrastructure that makes AI-skill claims meaningful at the point of hire.
“ Workers are not just leaving for higher pay. They are leaving because they believe their current trajectory runs into AI displacement, and because they believe their employer does not see it coming.”
The Verification Vacuum
76%
of US employed workers say misrepresenting AI-related skills puts businesses at risk. Yet only 39% believe employers can effectively verify these skills, and just 26% report that their employer has tested AI capabilities within the workforce. 43% believe most coworkers are exaggerating their AI skills. This report calls the resulting condition the Verification Vacuum: the structural absence that allows the AI Skills Bubble to keep inflating.
Percentage (%)
Percentage (%)
Workers see the Vacuum clearly
Despite widespread participation in the AI Skills Bubble documented in Sections 2 through 4, workers themselves are not in denial about its consequences. The same workforce that inflates AI claims also names the risk those inflated claims create for the organizations doing the hiring. The acknowledgment is not hypothetical. 27% of all workers say AI misrepresentation creates significant business risk, and another 49% say it creates some risk.
Employer-side verification has not caught up
64% of US workers say their employer has not attempted to verify their AI skills. 26% say their employer has tested or evaluated those skills. 10% are unsure. The asymmetry between the prevalence of inflated claims and the rarity of verification is the Verification Vacuum in concrete operational terms. It is also the structural condition that allows the inflation cycle in Section 4 to perpetuate. When 48% of inflators cite the absence of reliable verification as a reason they overstate, and 64% of all workers can confirm that absence from direct experience, the Vacuum is no longer a perception. It is a documented condition.
What the Verification Vacuum costs
The cost of the Vacuum distributes across both sides of the employment relationship. For workers, the cost is the post-hire stress of carrying an unverified claim into a role they cannot fully execute. For organizations, the cost is misallocation of work, delayed AI adoption, and the slow-burn risk of negligent hiring exposure when AI-skill claims that were never tested at hire-time later prove false in a high-stakes context. Neither side benefits from the current equilibrium. The 76% of workers who say misrepresentation creates business risk are also describing their own working conditions.
The Verification Vacuum
The structural condition in which AI-skill claims are made at scale while employer-side verification has not kept up. Workers know it. So do their colleagues. The vacuum is the condition that allows the inflation cycle to perpetuate.
64% of workers have never been tested by their employer. 48% of inflators cite the absence of reliable verification as a reason they overstate.
“ The same workforce that inflates AI claims also names the risk those claims create. The Verification Vacuum is not a perception problem. It is a structural one.”
What Workers Want: Transparency, Verification, Human Oversight
Workers are not opposed to verification. They are asking for it. 61% want human review of hiring decisions even when AI tools are used. 48% want testing of AI competencies. 47% want transparent explanations of how AI is used in hiring. 46% want clear communication about what will be verified. 42% want consistent standards applied across candidates.
The disclosure dividend
29% of US employed workers say they would be more honest about their qualifications if employers clearly communicated what would be independently verified. The figure represents a quantifiable behavioral lever, not a hypothetical preference. It signals that stronger, clearer screening processes could directly reduce AI skills misrepresentation. The implication for employer practice is specific: transparency is not only an ethical commitment. It is the highest-yield single intervention available for closing the Verification Vacuum documented in Section 7.
A complementary finding qualifies the ceiling. 13% of workers say they would still overstate their qualifications even if employers disclosed verification scope. The hardened fraction defines the upper limit of what disclosure alone can accomplish. Between the 29% who would respond and the 13% who would not lies the realistic operating range for transparency-driven screening reform.
Mapping the mandate to Compliance for Good
The Compliance for Good™ framework rests on three pillars: Transparent, Fair, and Protective Compliance. The 2026 data operates as the framework’s operating agenda for the automation era. Each pillar maps to a specific workforce mandate documented in this report.
| Pillar | What Workers Are Asking For | Anchor Statistic |
|---|---|---|
| Transparent Compliance | Communicate verification scope before candidates apply. Disclose what will be assessed. Make adverse-action reasoning visible. Provide candidates with the terms of their own screening. | 29% would present themselves more honestly if verification scope is disclosed |
| Fair Compliance | Consistent verification standards applied equally to all candidates. Dispute and reconsideration transparency. Individualized assessment for AI-skill claims that recognizes self-development context. | 42% cite consistent standards as a top trust-builder; 48% want direct testing of AI competencies |
| Protective Compliance | Verification infrastructure that can evaluate AI competency claims. Continuous monitoring for regulated roles. Protection against negligent hiring exposure from unverified AI skill claims. | 76% of workers themselves believe AI misrepresentation creates business risk |
Bridge to Trust in Hiring 2026
When the same screening-reform demands are put to US active job seekers and to US employed workers, the direction is identical. Job seekers demand transparency, fair standards, and human review at 74 to 82%. Employed workers demand the same things at 42 to 61%. The magnitude difference reflects proximity to the screening event. The directional convergence is what matters for policy: the workforce, taken as a whole, supports reform that makes screening more transparent, more consistent, and more human. Both populations cross the majority threshold on human review, at 81% and 61% respectively. This is the stakeholder consensus on which modern screening reform stands.
“ Workers are not opposed to verification. They are asking for it. The mandate for modern screening is coming from the people most affected by it. ”
The Bottom Line: Automation Anxiety Is Driving a New Trust Challenge
AI is creating the next wave of workforce disruption, and with it, a new kind of trust problem. Three structural conditions, all documented in this report, define the 2026 landscape: the AI Skills Bubble, the Double Distortion, and the Verification Vacuum. The Bubble is the inflated category of AI skill claim now appearing on resumes and in workplaces at industrial scale. The Double Distortion is the contradictory pattern of a workforce that simultaneously overstates AI capability and understates AI usefulness. The Vacuum is the absence of verification infrastructure that lets the first two persist. Together they describe a workforce caught between two pressures that point in opposite directions, operating inside an employer environment that cannot yet test the claims being made.
Employers operate inside the Verification Vacuum. The skill claims of the AI Skills Bubble have outpaced the infrastructure designed to evaluate them. 64% of workers have not had their AI skills tested. 39% believe their employer can effectively verify those skills. 43% believe their coworkers are exaggerating. The result is a growing distance between perceived and actual capability, and a measurable risk profile for both employees and organizations.
The path forward is not to screen more aggressively. It is to close the Verification Vacuum. The workforce itself has provided the mandate: transparency about what is being verified, consistent standards across candidates, human review of decisions in which AI tools are used, and direct testing of the new category of skill claim.
29% of workers would meaningfully change their behavior in response to disclosure alone, the highest-yield intervention available against the AI Skills Bubble. Compliance for Good in the automation era is a response to a workforce-generated mandate for screening that catches up with how workers actually behave in 2026.
The data does not argue that employers should screen more aggressively. It argues that employers should screen more credibly, and that the workers most affected by the resulting process have already given their consent for it to happen.
“ Compliance for Good in the automation era is not a brand position. It is a response to a workforce-generated mandate for screening that catches up with how workers actually behave in 2026. ”
Appendix. Demographic Cuts on Headline Metrics
By Generation
| Metric | Gen Z (n=132)† | Millennials (n=651) | Gen X (n=563) | Boomers (n=154)† |
|---|---|---|---|---|
| Threat perception (Q1 top-2) | 79% | 78% | 60% | 55% |
| AI fluency theater (any Q5) | 80% | 70% | 55% | 53% |
| Exaggerated AI experience | 50% | 38% | 21% | 19% |
| Volunteered for AI tasks unqualified | 30% | 24% | 12% | 8% |
| Explicitly lied about AI skills | 29% | 24% | 6% | 8% |
| Preemptive resume/profile change | 91% | 84% | 71% | 73% |
| AI resistance behavior (any Q7) | 87% | 83% | 78% | 77% |
| Currently looking at jobs (Q12 Yes) | 69% | 65% | 46% | 26% |
† Reported with small-sample caveat. Directional interpretation appropriate.
By Race and Ethnicity
| Metric | White (n=1,033) | Black (n=187) | Hispanic (n=133) | Asian (n=90)† |
|---|---|---|---|---|
| Threat perception (Q1 top-2) | 66% | 77% | 77% | 73% |
| AI fluency theater (any Q5) | 61% | 74% | 74% | 58% |
| Preemptive resume/profile change | 76% | 87% | 89% | 80% |
| Lists AI skills publicly (Q10) | 67% | 81% | 84% | 79% |
| AI fluency inflation (Q11 overstated) | 44% | 49% | 56% | 66% |
| Currently looking at jobs (Q12 Yes) | 49% | 66% | 69% | 61% |
† Asian subsample (n=90) reported as directional only.
Read: Hispanic and Black workers report systematically higher threat perception, AI fluency theater, preemptive behavior, and shadow job-search activity than White workers. The gradient is consistent across constructs. Workers of color carry a disproportionate share of the behavioral cost of AI anxiety.